PDF문서연방9주(상관순위차1).pdf

닫기

background image

상관관계(corelation):  변수간  상호관련성의  정도를  수치로  표현한  것

찰스  다윈(Charles  Darwin)의  사촌인  Francis  Galton은  특성의  개인차와  유전과의  관계(r)에 

관심을  가짐.    즉  부모의  키(stature)와  자녀의  키(신장)의  유전적  관계가  있는지  연구함.

그  후  이것을  Galton의  제자인  Karl  Pearson이  체계화시켜  정립시킨  것이  상관관계(r)임. 

1.    판매원의  수와  총판매액  사이의  단순  적률상관관계(Pearson  product-moment 

correlation)를  계산해보고  해석하라.    또  이것을  컴퓨터를  이용하여  구한  것과  비교해보시오.

--------------------------------------------------------------------------

X

Y

(X- 

)

  (X- 

)2  Y-

(Y- 

)2        Zx

    Zy

  ZxZy

--------------------------------------------------------------------------

15

135

-5

25

-65

4,225

-1.20

-1.19

1.43

18

    165

-2

-35

1,225

-0.48

-0.64

0.31

18

180

-2

4

-20

  400

-0.48

-0.37

0.18

22

230

  2

4

  30

  900

0.48

0.55

0.26

27

290

  7

49

  90

8,100

1.69

1.65

2.79

---------------------------------------------------------------------------

∑X=100

∑Y=1,000      ∑(X-

)2     

∑(Y-

)2 

     

∑ZxZy=4.97

  =

X/N

  =

Y/N

  =86

        =14,850

      =100/5=20

      =1,000/5=200

X의  분산=

(X-

)2/N=  86/5=17.2    X의  표준편차  =  

  =4.15

Y의  분산=

 

∑(Y-

)2  /N=  14,850/5=2,970    Y의  표준편차  =  

  =54.5

*  참고  Zx  =  (X- 

)/x표준편차  ;    Zy=  (Y- 

)/y표준편차 

r=sum  of  standardized  cross-product/number  of  observation

r= 

ZxZy/N=4.97/5=0.994  (판매원의  수와  총판매액  간에는  극히  높은  상관관계가  있다.)

        상관관계를  해석할  경우  교재  216쪽을  참고할  것.

2.    위의  두  변인들간의  관계를  Spearman의  순위차상관관계(rank-order  correlation)로       

                      계산해보고  해석하라.    또  이것을  컴퓨터를  이용하여  구한것과  비교해보시오.

X

Y

순위값(Xr)

순위값(Yr)

순위차(Xr-Yr)

      (Xr-Yr)2

 

15

135

    1

    1

    0

          0        

     

18

165

    2.5

    2

    0.5

        0.25           

18

180

    2.5

    3

  -0.5

        0.25

22

230

    4

    4

    0

          0

                27            290 

    5

    5

    0

          0

---------------------------------------------------------------------------

∑(Xr-Yr)2=0.50

Formula:     

Rho=  1-  [(6*sum  of  squared  ranking  difference)/n(n2-1)] 

            1-(6x0,50)/5(5

2-1)  =  1-(3.0/120)  =  0.97    매우  높다.   


background image

상관관계의  해석  (주교재  315~322,  부교재  216쪽을  참고할  것)

상관계수  r의  범위는  -1r+1  (교재에서  부등호의  방향이  잘못되었으니  주의할  것)

(상관계수는  –1보다  크거나  같고  +1보다  작거나  같다.  즉  절대값  1보다  클  수는  절대  없음)

   

상관계수  r의  절대값이

1.00이면  완전한  상관관계

0.90이면  매우  높은  상관관계

0.70~0.80이면  높은  상관관계

0.50~0.60이면  보통의  상관관계

0.30~0.40이면  약한  상관관계

0.10~0.20이면  매우  낮은  상관관계

0이면  상관관계가  전혀  없다고  말할  수  있다.

여가서  부호에  따라    +이면  정적인(양의)  상관관계

      -이면  부적인(음의)  상관관계에  있다고  하면  됨.

(a)  정의  상관

(b)음의  상관

(c)  무상관

            o  o

      Y

o    o

          Y   o    o

            Y         o  o    o  o

          o    o

        o    o

    o  o    o    o    o

    o    o

                o    o

        o    o    o  o

              o    o

        o    o

            o    o

       

          X

              X

              X